Software development requires extensive repetitive testing both of the first version and of each new version of the software that is produced throughout the development process. With each new version comes a repetitive series of tests to determine if the changes made to create the version have affected the software in an unintended manner. This testing is usually accomplished by a testing organization, which is separate from the software development organization, for example, in a different department.
In addition, software development is usually accomplished in modules that require complete testing as each new module is added to the software product. Thus, repetitive testing is also required to ensure that each new module, besides adding the functions it was designed to accomplish, will not detrimentally affect the functionality of the integrated software product. Software tests may be organized into test cases, so that different test cases test the functionality of different modules or sets of modules, or of different functionalities of the software.
Advances in technology have made it possible to record the keystroke input of an operator as the first version of software is tested. Subsequent versions are tested by playing back the previously recorded session. A system of this type is disclosed by Kerr et al. in U.S. Pat. No. 4,696,003,issued Sep. 22, 1987. Another relevant keystroke recording and playback system is the Advanced Test System For Software Test as disclosed in IBM Technical Disclosure Bulletin, Vol. 30, No. 9, Feb. 9, 1988. Still another recording and playback system is disclosed by Pope, et al. in U.S. Pat. No. 5,335,342, issued Aug. 2, 1994. In this system, a first version of interactive software is executed and all input signals, such as mouse activity and keystrokes, and output screens are recorded. When a new version of the software is created, the recorded input signals may be introduced to the software, and the new output screens compared to the recorded output screens.
Generally, software testing organizations use some type of record and playback functionality of input and/or output to produce one automated test script per test case. Each test script contains the data, execution logic and expected results for the test case. Within each script there may be common application functionality and logic.
Software testing organizations often use commercially available automated testing tools to test software. Record and playback test script generation is the most commonly demonstrated feature of commercially available automated testing tools. Record and playback functionality is one of the most intuitive methods to test software. Merely recording software input and playing it back tends to make the process of test automation appear very straightforward and easy. As the software application under test changes, however, it may also be necessary to change one or more of the test scripts. Maintaining a large number of test scripts that require frequent modification may become time consuming to the point of being overwhelming.
To help minimize the maintenance of test scripts, testing organizations have begun to use the record and playback functionality in conjunction with parameterized data in an effort to create more robust recorded scripts. Data is said to be “parameterized” when the automated test scripts read the test data values from an external source, such as a file, rather than having the values hard coded into the scripts. For example, pointer movements and selections that navigate through a series of screens may be “hard coded” into a script, while keyboard strokes (text data) may be read from a data structure or a data file. This method allows a variety of data scenarios to be tested by a single test script.
Nevertheless, even when using parameterized data with recorded scripts, many test cases have to be created to cover all of the target test cases. For example, one test script may be needed for each test case. Therefore, with each of the above methods it may become more difficult to maintain existing automated scripts as more test cases are automated. Maintenance of these scripts may become more complicated both during initial testing and later in regression testing. It would be desirable to provide an automated test system that allowed a single script to be used with many different test cases.